LanguageCrunch NLP Service docker image

Docker image: https://hub.docker.com/r/artpar/languagecrunch/

• Quickstart
• Endpoints
• Model Details
  • Sentiment
  • Entity extraction
  • Sentence type detection
  • Relation extraction
  • Word look up

Quickstart

Pull and run the Docker image, listening on port 8080:

$ docker run -it -p 8080:8080 artpar/languagecrunch

Example API call:

$ curl http://localhost:8080/nlp/parse?`echo -n "The new twitter is so weird. Seriously. Why is there a new twitter? What was wrong with the old one? Fix it now." | python -c "import urllib, sys; print(urllib.urlencode({'sentence': sys.stdin.read()}))"`

Endpoints

Sentence parse [Spacy]

GET http://localhost:8080/nlp/parse?sentence=<URL-encoded sentences>

Word lookup [Wordnet]

GET http://localhost:8080/nlp/word?word=ask&pos=v

Coreference resolution [neuralcoref]

GET http://localhost:8080/nlp/coref?sentence=<URL-encoded sentences>

Model Details

Sentiment

sentence: The new twitter is so weird. Seriously. Why is there a new twitter? What was wrong with the old one? Fix it now.

{
  "relations": [],
  "sentences": [
    {
      "sentence": "The new twitter is so weird. ",
      "sentence_type": "assertive",
      "sentiment": {
        "polarity": -0.18181818181818182,
        "subjectivity": 0.7272727272727273
      },
      "root": {
        "text": "is ",
        "orth": 2
      },
      "pos": [
        {
          "text": "The new twitter",
          "lemma": "the",
          "pos": "DET",
          "tag": "DT",
          "dep": "nsubj",
          .
          .
          .

Entity extraction

• PERSON
• NORP
• FACILITY
• ORG
• GPE
• LOC
• PRODUCT
• EVENT
• WORK_OF_ART
• LAW
• LANGUAGE
• DATE
• TIME
• PERCENT
• MONEY
• QUANTITY
• ORDINAL
• CARDINAL

Eg: Bill Gates, the founder of Microsoft, hosted a party last night

  "entities": [
    {
      "text": "Bill Gates",
      "label": "PERSON"
    },
    {
      "text": "Microsoft",
      "label": "ORG"
    },
    {
      "text": "last night",
      "label": "TIME"
    }
  ]
}

Sentence type detection

• assertive
• interrogative
• exclamatory
• negative

Relation extraction

Eg: Bill Gates, the founder of Microsoft, hosted a party last night

  "relations": [
    {
      "subject": "the founder",
      "object": "Microsoft",
      "relation": "ORG"
    }
  ],

Eg: Apple is looking at buying U.K. startup for $1 billion

 {
   subject: "N/A",
   object: "U.K. startup",
   relation: "GPE"
 },
 {
   subject: "buying",
   object: "$1 billion",
   relation: "MONEY"
 }
],

Word look up

• Category of word

  • Hypernyms - colour is a hypernym of red.

• Specific words of a category

  • Holonyms - red is a holonym of color

• Synonyms to match

• Examples

• Word frames ( how the word is used )

• Coreference resolution

• Pronouns/references to nouns

Eg: startle, verb

  "results": [
    {
      "definition": "to stimulate to action",
      "examples": [
        "..startled him awake",
        "galvanized into action"
      ],
      "lemma_names": [
        "startle",
        "galvanize",
        "galvanise"
      ],
      "hypernyms": [
        {
          "definition": "surprise greatly; knock someone's socks off",
          "examples": [
            "I was floored when I heard that I was promoted"
          ],
          "lemma_names": [
            "shock",
            "floor",
            "ball_over",
            "blow_out_of_the_water",
            "take_aback"
          ]
        }
      ],
      "lemmas": [
        {
          "frame_strings": [
            "Somebody startle somebody",
            "Something startle somebody",
            "Somebody startle somebody into V-ing something"
          ],

Contributors

List of contributors